The Inclusion Paradox: Why Europe's Employment Gap Is Not Closing

20 November 202616 min read
Share this article

InkludX Evidence Review, Paper 1 of 3. Europe's disability employment gap widened from 21.4 to 24.2 points in four years, and the biggest spenders do no better. We weigh the evidence on scale, trends, mental health and cost.

InkludX Evidence Review on Work Inclusion · Paper 1 of 3 · November 2026

In brief

  • The gap is not closing. In the EU, the employment gap between people with and without a disability was 21.4 percentage points in 2022 and 2023, 24.0 in 2024 and 24.2 in 2025 (Eurostat, ages 20–64).
  • Spending more does not buy a much smaller gap. Denmark and Norway spend more than twice the EU average on disability, yet their gaps in 2025 (23.5 and 23.8 points) were barely smaller than the EU's (24.2).
  • Exclusion is getting younger and is now mostly about mental and neurodevelopmental conditions. "A mental health epidemic" is too simple an explanation. The sudden jumps in the data line up with changes to benefit rules more often than with changes in health.
  • The cost is large. Public disability and sickness cash benefits alone come to 4.5% of GDP in Norway, and about €146.7bn a year across five north-European countries.
  • Our conclusion: Europe has built systems that protect income well and produce work poorly. The problem is how the money is spent, not how much.

How we weighed the evidence

This review brings together official statistics, OECD and EU analyses, government audits and evaluations, and peer-reviewed research. Every figure was read in its original source during this review. We did not take figures from memory or from secondary summaries.

We do not treat all evidence alike. We rank it in four tiers:

  1. Systematic reviews and randomised controlled trials (RCTs). The strongest answer to "does this intervention work?".
  2. Quasi-experiments on administrative register data. These are studies that use reforms, eligibility thresholds or cut-off dates as natural experiments. They are the strongest answer to "what did this policy do?".
  3. Official statistics. The strongest answer to "how big is the problem, and which way is it moving?". They cannot tell us why.
  4. Surveys of employers and workers, consultancy reports and modelled scenarios. Useful for forming hypotheses, but the weakest basis for conclusions.

We added three rules:

  • A well-designed study that finds no effect counts as much as one that finds an effect.
  • A result from one welfare system does not automatically carry over to another. A trial run in the United States is not evidence about Norway until it has been tested there.
  • Where countries define disability differently, we compare directions, not levels. The EU surveys use self-reported activity limitation. Germany uses registered severe disability. The UK uses the Equality Act definition.

Sources are at most three years old, unless they are the defining study on a question. Those older studies are dated in the text.


1. The gap is not closing

Eurostat's EU-SILC survey measures the employment rate of people aged 20–64 who report a long-standing activity limitation and compares it with everyone else. The gap was 21.4 percentage points in 2022 and 2023, then rose to 24.0 in 2024 and 24.2 in 2025. Much of the 2024 jump coincides with breaks in the survey series in several member states, including Italy and Spain, so it should not be read as a real deterioration. Even so, the gap has not narrowed in four years of data.

Labour-force participation shows the same picture. In 2024, 56.2% of people aged 20–64 in the EU with a limitation were working or looking for work, against 84.1% of those without one (EU-LFS).

The European Union's disability strategy, the UN Convention on the Rights of Persons with Disabilities and two decades of national activation reforms all set out to close this gap. On the official statistics, they have not.

2. The Nordic paradox

The countries that spend the most do not get the smallest gaps.

CountryEmployment gap 2025 (pp)Participation 2024, ages 20–64: limited vs not limited (%)Public spending on disability 2024 (% of GDP)
Norway23.857.2 vs 91.84.57
Denmark23.560.4 vs 93.15.04
Sweden22.671.8 vs 92.02.60
Netherlands24.364.2 vs 93.62.33
Germany23.064.8 vs 89.52.05
EU-2724.256.2 vs 84.11.91

Sources: Eurostat EU-SILC (hlth_dlm200), EU-LFS (lfsa_argaeddl) and ESSPROS (dsb_sprex01). "Spending on disability" covers both cash benefits and services such as care, rehabilitation and sheltered work. It does not include sickness benefits.

There are two reasons for the paradox.

  • The Nordic gaps are measured against a very high benchmark. Over nine in ten Norwegians and Danes aged 20–64 without a limitation are in the labour force. Disabled Norwegians participate at about the EU average.
  • High spending is not turning into employment. Denmark and Norway spend more than twice the EU average on disability, yet their gaps are only marginally smaller than the EU's.

The national registers show the scale behind these figures:

  • Norway. The OECD reports that sickness and disability spending is more than 8% of mainland GDP and around 22% of public spending (OECD Economic Survey of Norway 2026). At the end of 2025, 374,900 people received disability benefit (uføretrygd), 10.5% of those aged 18–67 (NAV). Counting the work assessment allowance (AAP) as well, about 500,000 people, roughly 14% of 18–66-year-olds, were on one or both in December 2023 (Perspektivmeldingen 2024).
  • Denmark. The number of people on disability pension (førtidspension) rose by 7,300 full-time equivalents in 2025, to 248,514. Counting every kind of public income support except student grants, 757,800 full-time-equivalent recipients, 19% of 16–66-year-olds, were supported in 2025, and the total has stayed close to 760,000 since 2020 (Statistics Denmark).
  • Netherlands. The number of incapacity benefits in payment rose from an annual average of 816,600 in 2021 to 865,000 in 2025 (CBS). More than 60,000 people a year now enter the long-term disability scheme (WIA), against about 35,000 when it was introduced. A government-commissioned interdepartmental review (IBO) concludes that the system has stalled and urgently needs far-reaching changes to both administration and legislation (Rijksoverheid, IBO "Werk aan de WIA").
  • Germany. Severely disabled people had an unemployment rate of 12.0% in 2025, against 7.6% for all unemployed people on the same reference base (BA, Blickpunkt Arbeitsmarkt 2025). In 2024, about a quarter (roughly 45,000) of the roughly 181,000 employers subject to the 5% employment quota filled none of their mandatory places (Bundesagentur für Arbeit).
  • United Kingdom. In Great Britain, 6.7% of people aged 18–66 were economically inactive because of long-term sickness at the end of 2025. That is 1.5 points above the pre-pandemic level (DWP). In England and Wales, 4 million 16–64-year-olds (one in ten) claim a disability or incapacity benefit, up from 2.8 million in 2019 (IFS).

Sweden is the instructive exception. The share of 19–65-year-olds on sickness compensation (sjukersättning) has fallen from 9.6% to about 3.5% over twenty years (Försäkringskassan). Yet in 2025 only 52% of Swedes aged 16–65 with a disability were employed, against 81% of the rest of the population (SCB). Sweden shows that the number of people on permanent benefits can be cut sharply. It does not show that cutting it moves people into work. Leaving benefits is not the same as being included in work. Paper 2 sets out the causal evidence behind this distinction.

3. Younger, and more often mental — but not one story

Across countries, people outside work because of their health are increasingly young, and increasingly have a mental or neurodevelopmental diagnosis:

  • Sweden. In December 2024, 53% of people on sickness compensation had a psychiatric diagnosis, up from 31% twenty years earlier (Försäkringskassan, 2025). In sick leave, psychiatric diagnoses account for about 30% of spells shorter than 28 days but more than 50% of spells lasting 180 days or more. Psychiatric spells last longer on average, so they make up a growing share the longer absences last (Försäkringskassan, 2026).
  • Netherlands. More than 40% of entries into the WIA are linked to psychological conditions (Rijksoverheid, IBO, 2025). Entries for psychological complaints were 29% higher in 2025 than in 2022, and the rise falls hardest on women under 40 (UWV, 2026).
  • United Kingdom. 44% of disability benefit claimants in England and Wales, 1.3 million people, claim mainly for a mental or behavioural condition (IFS). Among young people aged 18–24 who are not in employment, education or training (NEET), the share inactive because of sickness or disability rose from 12% in 2005 to 30% in 2025 (Resolution Foundation).
  • Denmark. Disability pension awards to people under 40 rose from 1,098 in 2014 to 6,270 in 2023 (Ankestyrelsen). 2014 was the low point after the 2013 reform, which tightened access and specifically aimed to keep people under 40 with mental disorders off disability pension, so the percentage rise is exaggerated.

Below the surface, the pattern differs from country to country:

  • Norway's young recipients rarely have depression or anxiety. 69.2% of disability benefit recipients aged 18–29 have a mental or behavioural primary diagnosis (NAV, 2026). But only about 3 in 100 had anxiety or depression at the end of 2023. Developmental disorders and intellectual disability are the most common diagnoses (NAV, 2025). The number of young recipients more than doubled, from 10,400 in 2013 to 21,800 in 2023 (Riksrevisjonen), and has been roughly flat since: 22,200 at the end of 2025.
  • Germany is the counter-example. The share of new disability pensions awarded for mental disorders was 42.6% in 2015 and 42.0% in 2024: high, but flat for a decade (Deutscher Bundestag, Drucksache 21/1792).

Four explanations, weighed

1. Mental health really is getting worse. The OECD finds that youth mental health indicators declined by an average of 3% to 16% a year between 2012 and 2022 in nine of the eleven countries with comparable data. In Britain, symptoms of common mental disorders among 18–24-year-olds rose from 24% in 2000 to 34% in 2021–22.

Our assessment: real, but gradual, and measured mostly through surveys.

2. Conditions are being recognised and diagnosed more. In Norway, growth among young recipients is linked above all to developmental, hyperkinetic and behavioural disorders (Helsedirektoratet). Asperger syndrome's share of new young recipients held at around 9–10% from 2017 to 2023, then rose to 12.7% in 2025 (NAV, 2026).

Our assessment: plausible, and it changes what support people need rather than whether they need it.

3. Work has changed. UWV breaks down the rise in WIA entries from 43,000 to 69,000 between 2018 and 2024 (UWV):

Share of the riseAttributed to
About 20%Changes in the size and make-up of the insured population (demography, economic cycle, higher pension age)
About 31%Assessment backlogs (advance payments counted as new awards, 22%) and two policy measures meant to reduce them (10%)
About 39%A higher probability of long-term illness: 22 points from the pandemic and long COVID, 16 points not yet explained but probably linked to an overheated labour market

Almost a third of the measured rise is therefore an artefact of how the benefit system is administered. UWV also reports that the rise in psychological drop-out since 2022 is concentrated in sectors with structurally high workloads due to staff shortages, such as care, education and public administration (UWV, 2026).

Our assessment: a real contributor, especially for women in high-pressure sectors, but in the Dutch data it accounts for at most the 16 unexplained points.

4. Benefit rules shape the numbers. The IFS finds that the post-pandemic rise in claims was almost unique to the UK. In almost all comparable countries, the share of working-age people claiming stayed close to pre-pandemic levels or fell, and Denmark was the only other country to see a significant rise. Two studies cited by NAV (Bragstad 2023; Lande 2019) find that the sharp rise in young disability recipients was mainly related to people approaching or reaching the maximum duration of the work assessment allowance (AAP). Danish awards fell sharply after the 2013 reform and have risen every year since 2014; municipalities say the 2018 legal clarification and the 2022 shortening of resource programmes added to the rise. In the Netherlands, almost a third of the rise in WIA entries comes from assessment backlogs and the measures taken to reduce them.

Our assessment: the strongest explanation for the sudden breaks in trend.

Our judgment. All four are at work. They differ in two ways that matter. Rising prevalence and better diagnosis change slowly, and governments cannot switch them off. The rules, and the design of work, change quickly and are under direct control. Germany's flat share alongside steep rises elsewhere suggests that no single shift in population health is behind the numbers. The sharpest movements in the data most often follow the rules. That makes institutional design the lever with the greatest potential, and the one most neglected.

4. What exclusion costs

Public budgets. The budget figures below are minimum costs: they leave out lost tax revenue, health care and employer-paid sick pay.

  • Cash benefits only. Public disability and sickness cash benefits are 4.5% of GDP in Norway, 3.4% in Denmark, 2.0% in Germany (2023), 1.8% in the Netherlands and 1.5% in Sweden. Together that is about €146.7bn a year (latest year, 2023–2024) across the five countries (InkludX fiscal cost profiles).
  • Norway's state budget. Payments of disability benefit, work assessment allowance and sickness benefit came to NOK 240bn in 2024, 12.5% of state budget spending (Riksrevisjonen, 2025).
  • The UK's trend. Working-age health-related benefit spending rose from £36bn in 2019–20 to £48bn in 2023–24, with £63bn forecast for 2028–29, all in 2024–25 prices (IFS).

The wider economy.

  • United Kingdom. The UK government estimates that working-age ill-health and disability that prevents work cost the UK economy £240–330bn in 2022 (DWP).
  • European Union. The OECD projects that mental ill-health will lower EU GDP by an average of 1.7% a year between 2025 and 2050, mainly through lower participation, absence and reduced productivity at work (OECD, 2026).

Benefits do their protective job. In 2024, 28.8% of disabled people in the EU were at risk of poverty or social exclusion, against 17.9% of others. Before any social transfers, counting pensions as transfers, 68.2% of disabled people aged 16 and over would be at risk of poverty; after transfers, 20.7% are (Eurostat). The case for reform is not a case for cutting protection. It is a case for adding work to protection.

The payoff from getting it right is the largest available. Norway's finance ministry modelled a scenario with fewer transitions to disability benefit: 20% fewer new AAP recipients under 35 who have no sickness-benefit entitlement, 20% less sickness absence lasting over three months (and 20% fewer moving from sickness benefit to AAP), and five percentage points more people moving from AAP into work. On its own, this would raise employment by just under 110,000 people by 2060 compared with the baseline, and increase fiscal room by NOK 54.1bn in 2033 (2024 kroner). That was the largest 2033 effect of any option in the ministry's comparison (Perspektivmeldingen 2024).

5. Conclusions

  1. Measure the outcome, not the caseload. The test of a work-inclusion system is how many disabled people are in work. How many people have left benefits is a different measure. Sweden and the Hungarian reassessment reform (Paper 2) show the two can move apart.
  2. Generosity is not the problem; passivity is. Spending protects people from poverty. What it fails to do in the Nordic countries and the Netherlands is lead into work.
  3. The systems were built for a different population. Disability insurance was largely built for people who lose work capacity late in a working life. The growing group today is young, has mental or neurodevelopmental conditions, and often has little work history. Rules and services designed for the first group fit the second poorly.
  4. Trends that policy made, policy can unmake. The sharpest breaks in the data most often follow changes to the rules. That is unwelcome, because some of the growth was avoidable. It is also hopeful, because it is within reach.

Paper 2 weighs the causal evidence on what moves people into work and keeps them there. Paper 3 sets out an agenda.


Sources

Tags

evidence-reviewdisability-employmentlabour-exclusionmental-healtheconomics
Share this article